Every model from the reference, in one sortable grid. Filter by modality, license posture, and local viability on Apple Silicon (M-series / MPS). Click any column header to sort. The two machine columns show local viability on the M5 Max (128GB unified) and the M5 Ultra (256GB, 80-core GPU): Native and Workable run there, Heavy is possible but slow, Cluster and Cloud-only do not run locally. The Ultra's extra memory and GPU push several models up a tier. The ComfyUI column tags how each model plugs into a node graph — Core (built-in / established local nodes), Custom node (community pack), API node (cloud call from inside ComfyUI), or None.Hover any maker or license badge for a deep-dive.
Snapshot: September 2026 — re-verify before productionLocal columns: M5 Max (128GB) & M5 Ultra (256GB, 80-core GPU) · MPS/ComfyUI